Conducting Survey Research in Strategic Management

1
Conducting Survey Research in Strategic Management

• ???????
• ???

2
Abstract

• Consider threats to the internal validity of
  field studies that utilize survey data.
• There are substantial threats to internal
  validity that fall into the general categories of
  sampling and measurement.
• The objective is to provide a road map for better
  use of survey methods.

3

• Primary data collected to specifically address a
  research objectives.
• Secondary data mean something other than what it
  was originally intended to mean.? demonstrate
  construct validity.

4
Sampling Issues

• Design an effective sampling plan.
• Sample representativeness
• Sample size
• Selection of key informants
• Strategies for securing high response rate and
  assessing non-response bias
• Issues in conducting online surveys

5
Sample Representativeness

• External validity
• Internal validity
• Internal validity is challenged by the
  relativity of strategy (Snow Hambrick, 1980,
  p.531)
• A reasonable way may focus on the early stages of
  research in a single corporation.
• Another alternative that provides controls for
  market-level influences is the single industry
  study.
• Multi-industry study the relationship truly
  robust

6
Sample Size

• Classical statistical significance testing
• Average sample size is 175 cases in SMJ that
  utilized primary data.
• Appropriate sample size ? statistical power
• Desired Type I error rate, effect size, desired
  power ? appropriate sample size
• multi-industry study large sample
• Single-industry study smaller sample
• Single-corporate study large enough to satisfy
  the requirements of the statistical techniques
  employed.

7
Selection of Key Informants

• The first challenge Identify respondents
  ability to report on the phenomenon under study.
• Key informants in strategy research are usually
  members of a top management team.
• Two approaches
• Use a general measure of informant competency
• Query the informant about his knowledge of the
  major issues covered in the study on a 1-5 scale.
• The second challenge improve the accuracy of key
  informant reports.

8
Strategies for Securing High Response Rate
Assessing Non-response Bias

• Two challenges
• Secure a high response rate
• Assess the degree to which non-respondents differ
  from respondents
• Of the 23 surveys reported in 20 articles in 2001
  2002 in SMJ, only 6 reported tests for
  non-response bias.
• Extrapolation method (Armstrong Overton, 1977)

9
Conducting Online Surveys

• Three major approaches
• E-mail surveys 1-10 days, the fastest and
  simplest of the three methods.
• HTML form surveys 3-15 days, the flexibility
• Downloadable interactive surveys 7-20 days
• Sampling error the most important issue when
  designing online surveys. (Ray Tabor, 2003)

10
Construct Measurement in Strategy Research

• The simplest and least risky approach use the
  existing measures.
• No guarantee the a valid measure will be
  developed.
• The validity of measures may change over time and
  with new situations.
• Should continue to provide evidence of
  reliability and of convergent and discriminant
  validity.

11
Reflective and Formative Indicators

• Two types of measures
• Reflective indicators
• CFA
• High relationship
• Formative indicators
• Diamantopoulos Winklhofer (2001)
• Represent a relatively small proportion of the
  measures used in strategy management

12
Scale Development

• A good measure must be both valid and reliable.
• Multi-item measures
• Two basic approaches to specifying the domain of
  the construct
• Deductive approaches
• Inductive approaches
• Define the construct literature reviews,
  critical incidents, panel of expert
• The wording of each statement should be precise.
• The number of items in the measure.
• Pretest
• Three tests in the scale purification process
  coefficient alpha, EFA, CFA

13
Reliability

• Coefficient alpha (0.7)
• A significant relationship between use of
  Likert-type scales and high reliability.
• Low alpha examine item-to-total correlations and
  eliminate items with the lowest correlations.

14
Validity

• Convergent and discriminant validity
• The multitrait-multimethod analysis
• CFA
• Three fit indices
• CFIgt0.9
• IFIgt0.9
• NNFIgt0.9
• RMSEAlt0.1
• Values for the average variance extractedgt0.5

15
Common Method Variance

• CMV (Common Method Variance)
• A good research design may reduce the likelihood
  of CMV.
• Post hoc factor analysis to find out a general
  factor.
• The Lindell and Whitney (2001) method for
  detecting CMV also corrects for CMV in the
  bivariate correlation. ? limited use in
  multivariate

16
Recent Example of Quality Survey Research

• Subramanium and Venkatraman (2001), Determinants
  of Transnational New Product Development
  Capability Testing the Influence of Transferring
  and Deploying Tacit Overseas Knowledge, SMJ, 22(4)

17
Summary

• Survey research is a valuable and valid strategy
  for conducting research on strategy-related
  issues.
• The starting point the objective(s)
• The end point report all the information